A common finding in the literature on satisfaction with medical care is that patients with worse health status, either physical or emotional, are less satisfied with their care. No research as been done that can shed light on the causes of this relationship, in spite of the evident implications of different explanations. The present research would test alternative causal explanations for the health status-satisfaction relationship, using structural equation modeling (LISREL). Comparable models will be estimated for four existing datasets with patient sample sizes ranging from 100 to over 600, each of which includes measures of the relevant constructs. Analyses of these datasets will provide independent replications of the main tests of causal explanations. Two main models will be estimated. The first is a direct or unmediated model, whereby the lower satisfaction of sicker patients is due to generalized attitudinal effects. The second model is a physician mediation model whereby the lower satisfaction of sicker patients is brought about by negative behavior on the part of health care providers, specifically physicians in the datasets to be analyzed; according to this hypothesis, physicians experience negative reactions to sicker patients, which are conveyed unwittingly to patients. These explanations differ in the locus of responsibility for the health status-satisfaction relationship; the direct model places responsibility on the patient, while the mediation model places responsibility on health care providers. The research will therefore guide our understanding of where corrective action can be taken.